Stochastic Behaviour Analysis of Adaptive Averaging Step-size Sign Normalised Hammerstein Spline Adaptive Filtering

Abstract

We introduce a sign algorithm based on the normalised least mean square with Hammerstein adaptive filtering using adaptive averaging step-size mechanism, which is derived by the minimised absolute a posteriori squared error. To improve the performance by reducing computational complexity, we suggest an adaptive averaging using energy of errors to update step-size variant. The analysis of convergence behaviour and mean square performance are derived. Experimental results reveal that the proposed algorithm can perform better than the least mean square approach based on the Hammerstein model of adaptive filtering.